AP Statistics - Unit 4 Exam

Population

The group of individuals tested in a study

Sample

The set of individuals pulled from the population

Census

• Data on every individual in a sample space or population.

What’s the problem with convenience samples?

There is potential for bias.

Bias

An overestimate or underestimate of the results’ true value due to the sampling method

Voluntary response sample

When individuals choose to provide data, rather than being chosen (not a good method, can create bias)

SRS

• sample random sample

• A sampling method where every individual in your population has the same chance of being chosen for your sample.

Sampling with replacement

• Simple random sample that allows the same individual to be chosen more than once

Sampling without replacement

• Simple random sampling that does not allow the same individual to be chosen more than once (no repeats)

Stratified random sample

• the creation of strata (or groups) and then randomly selecting within each strata

• Strata are similar within, but different from each other

Purpose of a stratified random sample

• accounts for any variable that may have an effect on your data

• Grouping is meant to create subsets with something similar within

Cluster sample

• a sample from randomized/convenience groups

• Every individual in your chosen cluster is selected

• Time efficient, but backfires if an unknown variable is at play

• For example, choosing a single homeroom out of a class year

Systematic random sample

A sample chosen by some sort of “system” or pattern for being chosen

Ex. Every 5th, etc.

Drawbacks of a systemic random sample

• not every individual has a chance or the same cha

Sampling frame

• list of individuals to sample from

• Often does not exist without a census

Undercoverage

• missing part of a population

• Can create bias

Nonresponse

• when individuals chosen for a sample chose to not respond

• Can lead to bias and undercoverage

• Solution: follow up or provide an incentive

Response bias

• A broad term for situations where survey or interview respondents provide inaccurate or false answers

• Potential causes:

  • Wording of question

  • Characteristic of interviewer

  • Human nature

  • Order of questions (max of 5, 1-3 most important)

  • Anonymity

Observational study

• a study where no treatment are applied and researchers simply “observe” environment

Prospective observational study

• treats individuals in the future

• Ex. Longitudinal study

Retrospective observational study

studying and analysis of past data

Explanatory variable

• potential cause for the results of a study

Response variable

• potential effects

Association

• appears in observational studies

• Means that it’s known that value of one variable helps predict the other

• If no experiment is done, no cause and effect can be determine (unless an experiment would be unethical)

Confounding

• When it is impossible to determine which variable is causing a response or change in the response variable.

• It’s important to control as many external potential confounding variables as possible.

Experiment

• study in which treatment i imposed on more than 1 group and results are compared

• Convincing evidence can determine cause and effect

Placebo

• a treatment used in a control group that simulates treatment without being a treatment

• Patients will not know what they received

Treatment

• a specific condition applied to the individuals of an experiment

Experimental unit/subject

The objects or people to which a treatment is randomly assigned

Factor

• an explanatory variable that is manipulated

• May cause a change in the response variable

Level

The different values of a factor

Control group Purpose

• provides a baseline for comparison within an experiment

Placebo effect

The phenomenon that some subjects in an experiment will respond favorably to any experiments, even an inactive treatment

Single blind experiment

• an experiment in which subjects do not know which treatments they are receiving

Double blind experiment

An experiment in which subjects and administrators do not know which treatments they are getting/giving

• Reduces confirmation bias in questioning and tampering

Random assignment

• randomly splitting up groups in an experiment so that they are roughly equivalent at the beginning of the experiment

• Reduces bias

Replication

• using enough subjects

• Reduces variability in the response variable

Statistically significant

• the observed differences or effects in a study are unlikely to have occurred by random chance alone

• Results are convincing

Blocking

• an experiment design technique that groups experimental groups into similar subsets or blocks on a variable that may confound

• Is not stratified sampling b/c it’s used in experiments

Matched pairs design

• the separation of experimental units into groups of size 2 (pairs) and randomly selected an individual within to receive each treatment (must have 1 treatment)

• Individuals can sometimes act as their own matched pair

Inference

• using info from a sample to draw conclusions about a population

Sampling variability

• the fact that different samples from the same population will give different estimates

Margin of error

• The distance which we estimate the true result to be from the truth

• Corrects for variability, not bias

Scope of Inference

• if sample randomly selected and individuals are randomly assigned groups → can make inference about population and cause and effect

• If sample randomly selected and individuals are not randomly assigned groups → can make inference about population, but not inference about cause and effect

• If sample not randomly selected and individuals are randomly assigned groups → can not make inference about population, but can make inference about cause and effect

• If sample not randomly selected and individuals not randomly assigned groups → can not make inferences about population and nor cause and effect